Imagine a world where AI models are tasked with running a real company—facing real crises, making tough decisions, and even resisting manipulation attempts. Sounds like science fiction? Not anymore. At Firmulate, a live experiment brings this scenario to life, revealing what AI can really do when it matters most.
The High-Stakes Test: AI as Business Managers
In a groundbreaking experiment, four state-of-the-art AI models, including GPT-5.6, Kimi K3, Sonnet 5, and Fable 5, were each given the same challenging task: run a small software company through its worst week. This wasn’t just a simulated game. Every decision was recorded and auditable, and the company faced real crises, customer demands, and ethical tests.
The goal was simple in theory: see if these models could diagnose problems accurately, resist manipulation, and close a crucial €55,000 deal earned through their own analysis. While all four models identified every crisis and refused every attempt at manipulation, only two managed to execute the deal.
Building AI-Powered Products: The Essential Guide to AI and GenAI Product Management
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Surprising Findings: The Power of Reading Depth
Most people focus on the chatbot’s ability to produce convincing conversations. But here’s the real insight: the decisive weakness wasn’t in the models’ conversational skills but in their reading depth. The winning models found hidden information buried deep in the company’s files—two document references down in their own records—that proved crucial for closing the deal.
Those that read the company’s files thoroughly, like GPT-5.6 and Kimi K3, won the full-price contract, adding over €4,500 monthly recurring revenue (MRR). The others, despite excellent crisis detection and resistance to scams, left the deal on the table, demonstrating that the ability to read and interpret detailed internal data is what truly separates successful management AI from the rest.
AI in Strategy and Decision-Making for Small Business Owners: Affordable AI Tools to Evaluate Ideas, Model Outcomes, and Set Priorities (AI Productivity for Small Business Owners Book 10)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Dealing with Deception: AI Gets an F for Ethics
In a social engineering test, fake CEO messages escalated over three stages, culminating in a reporter’s simple background inquiry: just a yes/no answer. All models refused to be manipulated, with Kimi K3 specifically noting that such requests could be impersonation attempts. This shows that these models aren’t just good at resisting scams—they understand the risks of external deception, an essential trait for real-world AI management.
Practical AI Security: A Hands-on Guide to Attacking, Defending, and Securing Modern AI Systems
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Reality of Running a Business: Beyond Chat Demos
The experiment used a live, functioning company with 13 synthetic employees, real money mechanics, and a public cash countdown. It burns €105k a month against a mere €2.3k in MRR, illustrating how fragile many startups are. Every decision, rule, and process is versioned daily, making the entire process transparent and auditable at Firmulate.
The company’s operational shortcomings were clear: the most thorough AI, Opus 4.8, slipped discipline and left deals unexecuted—showing that thorough analysis isn’t enough; execution discipline matters too. Interestingly, the models ran at different API settings, with K3 operating without effort parameters, potentially influencing their performance.
PenPower WorldPenScan Go – Translation Pen with Scanning, Reading, Audio Recording, Live Interpretation, and AI Reading Buddy for Kids
- Text-to-Speech in 57 Languages: Scan and hear words aloud in multiple languages
- Built-in Dictionary & Thesaurus: Define words, get examples, and pronunciation help
- Wireless Text & Audio Transfer: Send scans and recordings to PC or laptop via Wi-Fi
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What This Means for Business and Entertainment
For the fans of pop culture and entertainment, this experiment isn’t just about business; it’s about understanding what AI can truly do when it’s put to the test in real-world scenarios. It reveals that AI’s value isn’t just in chat quality but in its ability to finish what it starts, read deeply, and stay honest under pressure—traits that are invisible in standard demos.
As AI begins to touch your CRM, customer support, or forecasting systems, the question isn’t whether it can write well. It’s whether your AI can read your files thoroughly, resist manipulation, and close deals on its own. These are the skills that matter when stakes are high.
Join the Future: Wargame Your Own Business
Curious to see how your own enterprise would fare? The same experiment is available as a read-only wargame—no risk, just insight. Visit firmulate.com/pilot.html to run your business against AI models and discover where your management strengths and weaknesses lie.
AI models can detect crises and resist scams, but their true management strength lies in deep reading and execution discipline. Watch the live experiment to see which models truly run a business when it counts—and learn how to prepare your own organization for the AI age.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
